Divergence Color Histogram for Content-Based Image Retrieval
Combined With the color and spatial partitions of image, this paper puts forward divergence to describe the dispersion information of pixels in each annular subset, and all those divergences of annular subsets constitute the divergence color histogram. It is invariant with translation and rotation, and suitable as a basic feature of image. Finally, simulation results show the improvement of the divergence color histogram over the spatial color histogram are: 16% (annular), 27% (angular)and 26% (hybrid), so the divergence color histogram has a significant better retrieval performance.
Shaobo Zhong, Yimin Cheng and Xilong Qu
S. X. Xia et al., "Divergence Color Histogram for Content-Based Image Retrieval", Applied Mechanics and Materials, Vols. 50-51, pp. 639-643, 2011